Characterisation of a Novel Mouse Liver Aldo-Keto Reductase AKR7A5

FEBS 26272 FEBS Letters 523 (2002) 213^218

View metadata, citation and similar papers at core.ac.uk brought to you by CORE Characterisation of a novel mouse liver aldo-keto reductaseprovided AKR7A5 by Elsevier - Publisher Connector

Alison Hinshelwooda;b, Gail McGarvieb, Elizabeth Ellisa;c; aDepartment of Pharmaceutical Sciences, University of Strathclyde, 204 George Street, Glasgow G1 1XW, UK bSchool of Science and Technology, Bell College, Almada Street, Hamilton, Lanarkshire ML3 OJB, UK cDepartment of Bioscience, University of Strathclyde, 204 George Street, Glasgow G1 1XW, UK

Revised 12 June 2002; accepted 12 June 2002

First published online 25 June 2002

Edited by Stuart Ferguson

been identi¢ed in mouse liver to date. The distantly related Abstract We have characterised a novel aldo-keto reductase (AKR7A5) from mouse liver that is 78% identical to rat a£a- AKR7 family includes the rat liver a£atoxin B1 aldehyde re- toxin dialdehyde reductase AKR7A1 and 89% identical to hu- ductase or AFAR (AKR7A1), a second rat enzyme AKR7A4 man succinic semialdehyde (SSA) reductase AKR7A2. AKR7A5 [13,14] and two human homologues AKR7A2 [15] and can reduce 2-carboxybenzaldehyde (2-CBA) and SSA as well as AKR7A3 [16]. a range of aldehyde and diketone substrates. Western blots AKR7A1 was originally discovered as an inducible a£atox- show that it is expressed in liver, kidney, testis and brain, and in B1 dialdehyde reductase present in the livers of rats fed on at lower levels in skeletal muscle, spleen heart and lung. The diets containing the synthetic antioxidant ethoxyquin [17,18]. protein is not inducible in the liver bydietaryethoxyquin.Im- Subsequent studies revealed that AKR7A1 is also induced by munodepletion of AKR7A5 from liver extracts shows that it is a wide range of xenobiotics including butylated hydroxyani- one of the major liver 2-CBA reductases but that it is not the sole, coumarin, oltiprazand benzylisothiocyanate [19].Itis main SSA reductase in this tissue. ß 2002 Published byElse- vier Science B.V. on behalf of the Federation of European Bio- thought the gene is transcriptionally activated via an antioxi- chemical Societies. dant responsive element [20]. Following the characterisation of this inducible rat AKR, another related reductase has been Key words: Aldo-keto reductase; Succinic semialdehyde; found in rat (AKR7A4) [13] and two related human reduc- Mouse liver; Carbonyl metabolism tases (AKR7A2 and AKR7A3) [15,16] have also been iso- lated. Enzyme studies of these rat and human AKR7 enzymes have shown that they are capable of reducing a£atoxin 1. Introduction B1 aldehyde [15,16,21]. In addition, rat AKR7A1, human AKR7A2 and human AKR7A3 have all been shown to have Aldehydes and ketones form the reactive group in many substantial activity towards succinic semialdehyde (SSA), sug- endogenous and exogenous compounds, many of which are gesting that this compound, which is derived from Q-amino- toxic, mutagenic or carcinogenic [1]. The ability of mamma- butyrate (GABA), may represent an important physiological lian cells to metabolise these compounds depends on the pres- substrate for the AKR7A enzyme family [13,15,16,22]. Both ence of enzymes, including members of the aldo-keto reduc- rat AKR7A4 and human AKR7A2 proteins have been iden- tase (AKR) family [2], aldehyde dehydrogenases [3], alcohol ti¢ed in brain, suggesting a potential role in the biosynthesis dehydrogenases [4], carbonyl reductase [5], quinone reductase of Q-hydroxybutyrate (GHB), an important neurotransmitter, [6] and glutathione S-transferases (GSTs) [7]. from SSA [23]. In addition to exhibiting SSA reductase activ- Studies on the roles of AKR in carbonyl metabolism have ity, AKR7A family members are also able to metabolise assumed greater prominence in recent years, and this family 2-carboxybenzaldehyde (2-CBA), a compound that is struc- currently comprises 12 subfamilies (AKR1^AKR12) [8,9]. turally similar to SSA. Very few other AKRs can reduce Mouse liver AKRs represent a group of enzymes that contrib- 2-CBA, making this a relatively speci¢c substrate for ute to the metabolism of exogenous and endogenous com- AKR7A family members. Crystallisation of AKR7A1 has al- pounds. Enzymes present in adult mouse liver that have lowed the structure of this dimeric AKR to be determined, been cloned and characterised to date include AKR1C6 and has provided a structural basis for the enzyme’s ability (17-L-hydroxysteroid dehydrogenase) [10], AKR1C12 and to reduce 4-carbon acid aldehydes such as 2-CBA and SSA AKR1C13 [11], and AKR1E1 [12]. However, no members [24]. of the major aldehyde reductase AKR1A subfamily have A precise physiological function based on substrate speci- ¢city or tissue distribution alone has been di⁄cult to determine for any of the currently known AKR7A enzymes. At present it is not clear whether the rat AKR7A1 is the functional homologue of either human AKR7A2 or AKR7A3 and *Corresponding author. Fax: (44)-141-553 4124. there is no available evidence to suggest that either of the E-mail address: [email protected] (E. Ellis). human enzymes is inducible. Because of the potential useful- ness of mouse as a model for investigating in vivo carbonyl Abbreviations: AKR, aldo-keto reductase; SSA, succinic semialde- metabolism and detoxication, the cloning, expression, and hyde; GHB, Q-hydroxybutyrate; 2-CBA, 2-carboxybenzaldehyde; 4-NBA, 4-nitrobenzaldehyde; GST, glutathione S-transferase; 9,10- characterisation of a hitherto unrecognised AKR7A enzyme PQ, 9,10-phenanthrenequinone; NQO1, NAD(P)H:quinone oxidore- in the mouse is described, which it is hoped will lead to future ductase; GABA, Q-aminobutyrate in vivo functional analysis of the AKR7A gene family.

0014-5793 / 02 / $22.00 ß 2002 Published by Elsevier Science B.V. on behalf of the Federation of European Biochemical Societies. PII: S0014-5793(02)02982-4

FEBS 26272 5-7-02 214 A. Hinshelwood et al./FEBS Letters 523 (2002) 213^218

2. Materials and methods 2.4. Protein and enzyme assays Protein concentrations were measured using the method of Brad- 2.1. Cloning of mouse AKR7A5 cDNA ford [27] and standardised using bovine serum albumin, using a kit from Bio-Rad Laboratories Ltd. (Hemel Hempstead, UK). Aldehydes Comparison between rat AKR7A1 nucleotide sequences and other and ketone substrates were obtained from Sigma Chemical Co. cDNAs was carried out by BLAST searching [25] against the Gen- Bank database. A partial mouse AKR7A5 complementary DNA was (Poole, Dorset, UK) or from Aldrich Chemical Co. (Gillingham, Dor- initially ampli¢ed by RT-PCR using total mouse liver RNA as a set, UK). A£atoxin B1 dialdehyde was prepared using previously described methods, which allow the formation of the dialdehyde from source and oligos designed to the expressed sequence tag (EST) se- the dihydrodiol [28]. Aldehyde- and ketone-reducing activity was quences AA111743 and W50781 (MA3 ^ 5 -CCG.GAATTC.GCGT- P routinely measured with a Beckman DU650 UV single-beam record- CGCATGGATGCGAGTGCTGCTAGCG-3P and MA2 ^ 5P-CGG. GGTACC.GGACACTCGTGGGCGACC-3 ). This truncated cDNA ing spectrophotometer by following the initial rate of oxidation of P NADPH at 340 nm ( = 6270 M31 cm31) as described previously was cloned into pT7Blue3 to give pAH10 and was used to probe n [22]. Apparent K values for mouse AKR7A5 were determined by Northern blots. To determine the full-length cDNA sequence, m measuring the initial reaction rate over a range of substrate concen- SMART (switching mechanism at 5P end of RNA transcript) RACE (rapid ampli¢cation of cDNA ends) PCR was performed. Two gene trations and were calculated using Ultra¢t curve-¢tting software (Bio- soft, Cambridge, UK) using the Marquardt algorithm. speci¢c primers for the 3P and 5P RACE (GSP1 ^ 5P-CGTCCGGTC- CCAGTTAGAGACGTCTCTG-3 ; GSP2 ^ 5 -CCAAAGAAGCG- P P 2.5. Protein gels and Western blots GCCCACGGGTTGTTTCC-3P) were used in conjunction with primers speci¢c for oligonucleotides that had been incorporated into the SDS^PAGE was performed in 12% polyacrylamide resolving gels with the bu¡er system described by Laemmli [29]. For Western blot- 5 and 3 ends of cDNA derived from mouse liver mRNA. The full- P P ting, electrophoretic transfer of polypeptides from SDS^PAGE gels to length cDNA was then ampli¢ed from reverse-transcribed mouse liver RNA, using two oligonucleotides, one of which (mAFAR- nitrocellulose membranes was carried out using a Bio-Rad minigel FOR 5 -CGCTCCGGGACTTCGGTCGGGC-3 ) was designed to apparatus. Protein binding sites on membranes were blocked over- P P night in TBST (20 mM Tris^HCl/150 mM NaCl/0.01% (v/v) Tween) a third EST (accession no. AI893701) and the other (mAFARRev which contained 10% (w/v) skimmed milk. The blots were then incu- 5P-TGCTTTATTCAGACAGGA-3P) was designed to the 3P end from the sequence derived from the SMART RACE PCR. The se- bated for 1 h at room temperature with antisera against AKR7A5 or quence of the amplicon was veri¢ed and it was then used as template NAD(P)H:quinone oxidoreductase (NQO1) (at 1:2000 dilution in TBST). After washing, horseradish peroxidase-conjugated secondary for a second round of PCR with primers: mAFAR1 5 -CCG- P antibody (goat anti-rabbit IgG, at 1:3000 dilution) in TBST was GAATTCCAT.ATGTCCCGGCCTCCGCCACCCCGC-3P (forward) and mAFAR2 5 -CGGCTCGAG.CTATCTGAAGTAGGTTGGGA- added. The antibody complexes were detected using enhanced chemi- P luminescence (ECL; Amersham). CA-3P (reverse) which included 5P EcoRI and NdeI sites and a 3P XhoI site. This full-length mouse AKR7 cDNA was blunt cloned into the 2.6. Animals and preparation of tissue extracts multiple cloning site of pT7Blue3 (Novagen) between the EcoRI sites resulting in pAH11. The DNA insert in pAH11 was sequenced to Eight week old male CD-1 mice were fed a basic diet of powdered verify ¢delity of ampli¢cation using dye terminators and Amplitaq food with arachis oil, or the basic diet/arachis oil plus 0.5% (w/w) ethoxyquin, butylated hydroxyanisole, coumarin or benzyl isothiocya- FS DNA polymerase and separated on an ABI (Applied Biosystems) nate for 7 days immediately before sacri¢ce. Upon removal, the or- 373A automated sequencer. Multiple sequence alignments of AKR7A 3 family members were carried out using Clustal-X [26]. gans from the CD-1 mice were snap-frozen and stored at 70‡C until use. To prepare the cell extracts approximately 100 mg of each tissue 2.2. Bacterial expression and isolation of recombinant AKR7A enzyme was taken and homogenised in 20 mM NaPO4 bu¡er (pH 7.0). After centrifugation to remove cell debris, supernatant fractions were stored The coding region for mouse AKR7A5 was excised from pAH11 by 3 digestion with NdeI and XhoI, and ligated into the NdeI and XhoI at 80‡C. sites of pET15b expression vector. The resulting expression construct AKR7A5 was immunodepleted from total liver extracts as follows: a 100 Wl aliquot of preswollen protein A-Sepharose beads (approxi- pAH12 was used to transform Escherichia coli BL21 pLysS. Expres- mately 50% slurry) was added to 200 Wl of antisera (either against sion of recombinant His-tagged AKR7A5 protein was induced for AKR7A5 or LDH). After incubation at 4‡C for 1 h, portions of 1.5 h at 37‡C by the addition of 0.5 mM isopropyl L-D-thiogalactoside to mid-exponential phase transformed BL21 pLysS cells. Cells were extracts containing equivalent amounts of AKR7A5 (as estimated from quantitative Western blots) diluted into 200 Wl of ice-cold harvested by centrifugation, frozen at 370‡C, and lysed using a son- 10 mM sodium phosphate bu¡er (pH 8.0) were added to each of icator in bu¡er A (20 mM sodium phosphate, 500 mM NaCl pH 7.4). Lysates were ¢ltered through a 0.45 Wm Whatman polysulphone disc, the protein^antibody mixtures; after incubation at 4‡C for 2 h, the and loaded onto a 5 ml HiTrap Chelating a⁄nity column (Pharmacia beads were removed by centrifugation. Complete removal of AKR7A5 from the extracts was veri¢ed by Western blot analysis of Biosystems Ltd., Milton Keynes, Herts, UK), pre-equilibrated with the supernatants (data not shown). The resulting AKR7A5- and bu¡er A. After washing the column, recombinant polyhistidine-tagged protein was recovered by elution with 200 mM imidazole in bu¡er A. LDH-depleted supernatants were then assayed for carbonyl-reducing The eluted recombinant protein was then passed through a G25 gel activity. ¢ltration system column, equilibrated with 20 mM sodium phosphate bu¡er (pH 6.6) to remove the imidazole and salt. 10% (v/v) glycerol 3. Results and discussion was added and puri¢ed enzyme to be stored at 370‡C. Removal of the His-tag was achieved using human thrombin (Sigma, Poole, UK). 3.1. Identi¢cation and cloning of mouse AKR7A5 Molecular weight estimation was performed using a DYNA-PRO 801 dynamic light scattering/molecular sizing instrument (Protein Solu- The rat AKR7A1 sequence was used to carry out a BLAST tions, Buckingham, UK). search against the GenBank database. This identi¢ed several mouse ESTs whose predicted gene products showed extensive 2.3. Antibodies to AKR7A5 similarity to AKR7A1 at the amino acid level. Two of these Puri¢ed mouse AKR7A5 (approximately 100 Wg in 1 ml 10 mM sodium phosphate bu¡er; pH 7) was emulsi¢ed with an equal volume sequences (AA111743 and AI893701) were derived from adult of Freund’s complete adjuvant and injected subcutaneously at four kidney cDNA whereas the third was embryonic in origin. separate sites on the back of two female New Zealand White rabbits. These ESTs were identical in their overlapping regions, sug- After 14 days each rabbit was re-inoculated with 100 Wg of the orig- gesting that they were derived from the same gene (Fig. 1). inal immunogen in incomplete Freund’s adjuvant and this was fol- SMART RACE PCR was used in an attempt to determine lowed by a ¢nal inoculation 10 days later, this time with 100 Wgof immunogen in 0.9% saline. After a further 10 days the animals were whether the sequence represented the full-length mouse killed and bled. The serum obtained was stored at 320‡C, in the AKR7A-related sequence. Sequencing of these amplicons re- presence of 0.1% sodium azide until required. vealed additional sequence at the 3P end, including the polyA

FEBS 26272 5-7-02 A. Hinshelwood et al./FEBS Letters 523 (2002) 213^218 215

20% identity with other mouse AKRs in the major AKR1 family, and also shows some similarity in structure and sequence to the voltage-gated Shaker-related Kþ channel L-sub- unit AKR6 family. The four amino acids forming the catalytic tetrad shown to be involved in catalysis in other AKRs [30] are conserved in mouse AKR7A (Tyr-56, His-130, Lys-84 and Asp-51), suggesting that the mouse sequence is likely to en- Fig. 1. Alignment of mouse EST showing similarity to rat AKR7A1 code an active enzyme. cDNA. Three ESTs were identi¢ed by using the AKR7A1 cDNA sequence to perform BLAST search of the GenBank database. Ad- 3.2. Catalytic properties of mouse AKR7A5 ditional 3P sequence was derived from 3P RACE of mouse liver To determine whether the mouse AKR7A-related protein mRNA. The full-length sequence has been deposited in the EMBL database. was catalytically active, the coding region was cloned into a bacterial expression vector, and the protein puri¢ed as described previously for rat AKR7A1 [18]. Removal of the tail. However, no additional sequence could be determined at N-terminal His-tag using thrombin resulted in a protein which the 5P end beyond that which was present in EST AI893701. co-migrated with an AKR7-related protein present in mouse To con¢rm that this sequence represents a bone ¢de cDNA liver extracts (Fig. 3A), indicating that it represents the full- that is expressed in mouse liver, RT-PCR was performed from length protein that is expressed in liver. Dynamic light scatter- mouse liver total RNA using primers to the 5P and 3P ends of ing analysis showed the puri¢ed protein to be monodispersed, the proposed sequence. A PCR product of 1.01 kb was iso- with a hydrodynamic radius of 3.8 nm, giving it an estimated lated, cloned and the full-length sequence was con¢rmed by molecular weight of 77 kDa. This means that the puri¢ed DNA sequencing. The consensus sequence resulted in an open recombinant protein is a dimer, a feature which it shares reading frame of 1017 bp, which would translate to a poly- with rat AKR7A1 and AKR7A4 [13,24]. peptide of 338 residues with an estimated molecular mass of Enzyme assays were performed which showed that the pu- 37 662 Da. Comparison of this putative mouse AKR7A and ri¢ed recombinant mouse protein was able to reduce the mod- other already identi¢ed AKR7A enzymes (Fig. 2) showed these proteins are highly related. At the amino acid level the putative mouse AKR7A is 78% identical to the ethoxyquin- inducible a£atoxin dialdehyde reductase (AFAR; AKR7A1) from rat liver and 88% identical to both human AKR7A2 and AKR7A3. From this sequence similarity alone it might be expected to perform a similar role to one or more of these enzymes. In common with its close relatives, it shares less than

Fig. 3. Detection of AKR7A5 in mouse tissues. Detection of recombinant AKR7A5 and an AKR7A5-related band in mouse tissues was carried out by separating puri¢ed protein and tissue extracts by Fig. 2. Multiple sequence alignment of full-length mouse AKR7A5 SDS^PAGE and subjecting to Western blotting. The membranes with rat and human AKR7A sequences. Clustal-X multiple sequence were probed with antisera raised to AKR7A5 and detected using alignment between the amino acid sequences of AKR7A5 and other ECL. A: 1, Recombinant AKR7A5; 2, recombinant AKR7A5 after AKR7A. *, identical amino acids; : and ., conserved amino acids; cleavage of the His-tag with thrombin; 3, 10 Wg mouse liver extract. +, amino acids corresponding to the catalytic tetrad. B: 10 Wg extract of the mouse tissues indicated.

FEBS 26272 5-7-02 216 A. Hinshelwood et al./FEBS Letters 523 (2002) 213^218 el substrate 4-nitrobenzaldehyde (4-NBA). It was therefore Table 2 given the name AKR7A5 as it represents the ¢fth active mem- Comparison of catalytic e⁄ciency of recombinant mouse AKR7A5, rat AKR7A1 and human AKR7A2 ber of the subfamily to be characterised. Like the rat and 31 31 human AKR7A enzymes, the mouse enzyme can use both Kcat/Km (min M ) NADH and NADPH as cofactor (Table 1). Using 4-NBA Mouse AKR7A5 Rat AKR7A1 Human AKR7A2 as a substrate, and NADPH as a cofactor, the optimum pH 4-NBA 5.34U104 9.55U104 1.1U104 for the mouse AKR7A5 was determined and found to be pH 9,10-PQ 7.69U106 4.13U106 1.48U107 U 6 U 7 U 7 6.6, but the enzyme can function over a broad range (pH 5.0^ 2-CBA 4.86 10 1.41 10 1.14 10 SSA 4.00U106 1.06U106 3.02U106 9.0). AKR7A5 also demonstrates a broad range of substrate 2-NBA 1.62U105 1.4U105 6.23U104 speci¢cities for aldehydes and diketones, including drugs, The activities of AKR7A5, AKR7A1 and AKR7A2 towards car- toxic aldehydes, and metabolites. bonyl-containing compounds were determined at 25‡C as described The apparent Michaelis constants of the recombinant in Section 2 using NADPH as cofactor. Apparent Km and Kcat val- mouse AKR7A5 for some relevant substrates were determined ues were estimated from the initial velocities measured over a range (Table 1). Mouse AKR7A5 can reduce the dialdehyde form of of substrate concentrations using the Ultra¢t curve-¢tting programme. AFB1-dihydrodiol (pH 8.5), although it has a lower a⁄nity (Km 90 WM) compared with that determined previously for the rat and human enzymes (9 WM) [16,21]. This is particularly 2-CBA, which has been identi¢ed as a relatively speci¢c sub- salient because it had been thought that an a£atoxin-metab- strate for AKR7A enzymes (Table 2). Mouse AKR7A5 had a olising aldehyde reductase would not be important for detox- similar a⁄nity for 4-NBA (Km = 870 WM) as rat AKR7A1 ication of this compound in mouse because of the high con- (Km = 520 WM), whereas human AKR7A2 had signi¢cantly stitutive presence of a speci¢c alpha class GST in this species. lower a⁄nity (Km = 6630 WM). All of the AKR7A enzymes This GST is able to conjugate the a£atoxin 8,9-epoxide and tested reduced the bulky hydrophobic dicarbonyl compound was thought to be responsible for conferring the high level of 9,10-PQ, but the mouse and human enzymes have higher af- resistance to a£atoxin B1 seen in mouse [31]. A similar GST is ¢nity and catalytic activity. SSA, which has been shown pre- inducible in rat [32]. In humans, the AKR7A enzymes are viously to be reduced by human AKR7A2 [15], is also reduced considered to assume greater importance [21]. The catalytic e⁄ciently by mouse AKR7A5, whereas rat AKR7A1 has a e⁄ciency that we observe suggests that AKR7A5 may also much lower a⁄nity and turn over rate than the mouse or make a contribution to a£atoxin resistance in the mouse. human enzyme. AKR7A5 is also capable of reducing trans, trans-muconaldehyde which is a toxic metabolite of benzene and has been 3.3. Tissue distribution of mouse AKR7A5 linked to the occurrence of leukaemia. The endogenous mono- Antibodies raised against mouse AKR7A5 were used to amine oxidase inhibitor isatin has been shown previously to probe Western blots of various mouse tissue protein extracts be a good substrate for human AKR7A2 [15], and likewise, it (Fig. 3B). Mouse AKR7A5 antibodies reacted with a protein is also a good substrate for mouse AKR7A5 with a Km of 109 band in most tissues tested; however, levels are highest in WM. However, mouse AKR7A5 has very weak a⁄nity for liver, kidney, testis and brain, and lower levels were found some dicarbonyls such as diacetyl (Km 58 mM), which marks in the heart, spleen, and skeletal muscle. This pattern of ex- it as clearly di¡erent from rat AKR7A1. The a⁄nity of the pression di¡ers from rat AKR7A1, as the rat protein is only enzyme for NADH is signi¢cantly lower than that observed detectable at low levels in the livers of control rats and is for NADPH (Table 1) when using 4-NBA as substrate. present additionally only in testis, kidney and small intestine The kinetic constants of the mouse, rat and human AKR7A [18]. The tissue distribution of AKR7A5 is more similar to enzymes were compared for some model AKR substrates: that seen for the human AKR7A2 protein which is expressed SSA, 9,10-phenanthrenequinone (9,10-PQ), 4-NBA and also in most tissues and is detected at high levels in liver, kidney,

Table 1 Apparent kinetic constants for mouse AKR7A5 for aldehyde and diketone substrates Substrate Mouse AKR7A5 31 31 31 Km (WM) Kcat (min ) Kcat=Km (min M ) 4-NBA 870 þ 500 46 þ 6 5.34U104 9,10-PQ 8.0 þ 4.0 62 þ 6 7.69U106 2-CBA 16 þ 3 78 þ 17 4.86U106 SSA 20 þ 10 80 þ 12 4.02U106 5 AFB1-dihydrodiol (pH 8.5) 90 þ 30 34 þ 7 3.77U10 Isatin 109 þ 26 54 þ 3 4.99U105 2-NBA 460 þ 140 75 þ 11 1.63U105 Methyl glyoxal 3 980 þ 940 47 þ 3 1.18U104 Camphorquinone 4 990 þ 490 66 þ 3 1.32U104 Trans, trans-muconaldehyde 7 190 þ 5 990 94 þ 41 1.31U104 Diacetyl 58 420 þ 16 410 114 þ 1 1.96U103 NADPH 1.62 þ 0.01 26 þ 3 1.58U107 NADH 1 470 þ 23 115 þ 101 7.86U104 The activity of AKR7A5 towards carbonyl-containing compounds was determined at 25‡C as described in Section 2 using NADPH as cofactor. Apparent Km and Kcat values were estimated from the initial velocities measured over a range of substrate concentrations using the Ultra¢t curve-¢tting programme. The values shown represent mean þ S.E.M. The Km and Kcat for NADPH and NADH were estimated using 4-NBA as substrate.

FEBS 26272 5-7-02 A. Hinshelwood et al./FEBS Letters 523 (2002) 213^218 217

liver. In many tissues, SSA is produced from GABA by GABA-transaminase and is metabolised by SSA dehydrogenase to succinic acid. However, SSA can also be reduced by SSA reductase in some tissues to GHB, particularly where SSA dehydrogenase is not ubiquitously expressed. GABA- transaminase is widely distributed in non-neural tissues, including liver, kidney and testis whereas SSA dehydrogenase is absent from some of those tissues such as the testis. A role for AKR7A2 in the reduction of SSA has been supported by Fig. 4. Induction of mouse liver enzymes by chemoprotectors. Por- the puri¢cation of AKR7A2 from human brain as a major W tions (10 g) of hepatic extract from control male mice and from SSA reductase [23]. Similarly, AKR7A-related proteins are male mice that had been fed ethoxyquin, tert-butyl hydroquinone, coumarin, indole-3-carbinol or benzyl isothiocyanate were separated expressed in rat brain [13,35]. However, it has been demon- by electrophoresis on a single SDS^PAGE gel. After blotting, the strated that AKR7A2 does not constitute the major SSA re- nitrocellulose membrane was probed with antisera raised to ductase in human liver [15], as the major peak of SSA reduc- AKR7A5 or NQO1. Cross reacting polypeptides were visualised by tase activity does not copurify with AKR7A2. A similar ECL. 1, Control diet; 2, coumarin; 3, ethoxyquin; 4, butylated hydroxyanisole; 5, benzyl isothiocyanate. observation was seen in rat liver where neither of the two rat enzymes AKR7A1 or AKR7A4 was present in the major SSA reductase peak [13]. Our results suggest that AKR7A5 is testis and, importantly, brain [33]. The narrow tissue distribu- not the major SSA reductase in liver, which is concordant tion of the rat enzyme suggested that it has a specialised tissue with it paralleling the function of human AKR7A2. Presum- speci¢c role, possibly in detoxication. By contrast, the more ably another AKR such as an aldehyde reductase of the widespread rat AKR7A4 and human AKR7A2 tissue distri- AKR1 family contributes the majority of SSA reductase ac- bution suggests that these enzymes may perform a ‘house- tivity in the liver. Further work is required to establish keeping’ role, possibly in metabolism [13,15]. Overall, the whether AKR7A5 is a major SSA reductase in mouse brain, mouse enzyme distribution seems to ¢t more closely with particularly if other AKRs are not present in this tissue. the latter two enzymes, suggesting that it too performs a Immunodepletion experiments did, however, reveal that house-keeping role. there is a signi¢cant decrease (40%) in hepatic 2-CBA reductase activity following removal of AKR7A5 from the extracts. 3.4. Induction of mouse liver enzymes by chemoprotectors 2-CBA is a carbonyl compound that is of interest to the An unusual feature of the rat AKR7A1 is that its expres- pharmaceutical industry because it is used to mask drugs sion in liver is inducible by a variety of chemopreventive com- and enhance their delivery to target sites [36,37]. It has been pounds, including synthetic antioxidants such as ethoxyquin, previously reported that AKR7A2 is the principal 2-CBA re- butylated hydroxyanisole and oltipraz [34]. It is therefore con- ductase in human liver [15], and our data suggest that this role sidered to be a key detoxication enzyme that is part of an may also be shared by AKR7A5 in mouse liver. Immunode- adaptive stress response to chemical and oxidative stress pleted extracts also showed that AKR7A5 is responsible for [20]. In contrast, rat AKR7A4 is found to be constitutively approximately 15% of the total hepatic 4-NBA reductase ac- expressed in the liver [13]. To determine whether AKR7A5 tivity (Fig. 5), but did not show any decrease in 9,10-PQ was inducible, mice were fed on diets containing ethoxyquin, reductase activity, suggesting that it is not the major liver butylated hydroxyanisole, coumarin, and benzyl isothiocya- enzyme responsible for the reduction of this compound nate for 7 days and the livers examined for increased expres- (Fig. 5). sion of AKR7A5 and NQO1, an enzyme previously shown to be inducible by these compounds. Fig. 4 shows that although the treatments lead to the induction of NQO1, there is no increase in the levels of AKR7A5. The lack of inducibility of AKR7A5 suggests that AKR7A5 is assuming a role similar to the constitutive rat AKR7A4. It is possible that the inducibility of AKR7A1 is a feature unique to the rat. Neither of the human enzymes AKR7A2 and AKR7A3 has been shown to be inducible in the liver, though there does appear to be some interindividual variation in expression levels and activities [21,33].

3.5. Contribution of mouse AKR7A5 to liver carbonyl-reducing activity A low Km for SSA suggested that the mouse AKR7A5 could function as a SSA reductase in vivo, as has been pro- Fig. 5. Carbonyl-reducing activities in hepatic cytosols immunode- posed for human AKR7A2 [15]. To assess the role of mouse pleted for AKR7A5. Extracts from the livers of three mice that had AKR7A5 in the metabolism of SSA in liver, we depleted liver been fed either control diet or control diet plus ethoxyquin were in- extracts using antibodies raised to AKR7A5 (Fig. 5). This cubated with antisera raised against AKR7A5 (open bar) or LDH (speckled bar) (as a control) and protein A-Sepharose beads. After showed that SSA reductase activity in mouse liver does not removal of the beads by centrifugation, the remaining activities in decrease when mouse AKR7A5 is removed, which indicates the immunodepleted samples were assayed for the substrates indi- that it is not the primary SSA reductase enzyme in mouse cated. Data represent the mean of three experiments (n = 3).

FEBS 26272 5-7-02 218 A. Hinshelwood et al./FEBS Letters 523 (2002) 213^218

4. Conclusions [11] Ikeda, S., Okuda-Ashitaka, E., Masu, Y., Suzuki, T., Watanabe, K., Nakao, M., Shingu, K. and Ito, S. (1999) FEBS Lett. 459, 433^437. The data we present point to the mouse AKR7A5 being the [12] Bohren, K., Barski, O. and Gabbay, K. (1996) in: Enzymology functional orthologue of human AKR7A2 and rat AKR7A4. and Molecular Biology of Carbonyl Metabolism 6 (Weiner, H., This conclusion is based on the fact that it shares similar Ed.), Plenum Press, New York. enzyme kinetic properties, tissue distribution pattern and [13] Kelly, V.P., Ireland, L.S., Ellis, E.M. and Hayes, J.D. (2000) Biochem. J. 348, 389^400. does not appear to be inducible. In addition, we have shown [14] Nishi, N., Shoji, H., Miyanaka, H. and Nakamura, T. (2000) that it, like the human AKR7A2 and rat AKR7A4 enzymes, Endocrinology 141, 3194^3199. is not the major SSA reductase in liver, but that it is likely to [15] Ireland, L., Harrison, D., Neal, G. and Hayes, J. (1998) Bio- contribute to 2-CBA reduction in that tissue. It may also play chem. J. 332, 21^34. a role in a£atoxin B dialdehyde reduction in mouse liver, [16] Knight, L.P., Primiano, T., Groopman, J.D., Kensler, T.W. and 1 Sutter, T.R. (1999) Carcinogenesis 20, 1215^1223. thus contributing to constitutive resistance to the a£atoxin. [17] Hayes, J., Judah, D. and Neal, G. (1993) Cancer Res. 53, 3887^ A role for AKR7A5 as a SSA reductase in brain remains to 3894. be established. [18] Ellis, E., Judah, D., Neal, G. and Hayes, J. (1993) Proc. Natl. The characterisation of the enzyme described here will pave Acad. Sci. USA 90, 10350^10354. [19] Ellis, E., Judah, D., Neal, G., OConnor, T. and Hayes, J. (1996) the way for its function in vivo to be investigated through Cancer Res. 56, 2758^2766. the generation of a mouse line in which the AKR7A5 gene [20] Hayes, J., Ellis, E., Neal, G., Harrison, D. and Manson, M. has been knocked out. This will enable future investigations (1999) Biochem. Soc. Symp. 64, 141^168. into its role in detoxication and drug metabolism, as well [21] Guengerich, F.P., Cai, H., McMahon, M., Hayes, J.D., Sutter, as its potential role in the metabolism of aldehydes in the T.R., Groopman, J.D., Deng, Z. and Harris, T.M. (2001) Chem. Res. Toxicol. 14, 727^737. brain. [22] Ellis, E. and Hayes, J. (1995) Biochem. J. 312, 535^541. [23] Schaller, M., Scha¡hauser, M., Sans, N. and Wermuth, B. (1999) Acknowledgements: We are grateful to Prof. Iain Hunter for use of Eur. J. Biochem. 265, 1056^1060. the spectrophotometer and to Dr. Adrian Lapthorn for use of the [24] Kozma, E., Brown, E., Ellis, E.M. and Lapthorn, A. (2002) dynamic light scattering machine. We thank Prof. Gisela Witzfor J. Biol. Chem. 277, 16285^16293. supplying trans, trans-muconaldehyde. A.H. was supported by a [25] Altschul, S.F., Gish, W., Miller, W., Myers, E.W. and Lipman, Bell College Postgraduate Studentship. D.J. (1990) J. Mol. Biol. 215, 403^410. [26] Jeanmougin, F., Thompson, J.D., Gouy, M., Higgins, D.G. and Gibson, T.J. (1998) Trends Biochem. Sci. 23, 403^405. [27] Bradford, M.M. (1976) Anal. Biochem. 72, 248^254. References [28] Johnson, W., Harris, T. and Guengerich, F. (1996) J. Am. Chem. Soc. 118, 8213^8220. [1] Sladek, N., Manthey, C., Maki, P., Zhang, Z. and Landkamer, [29] Laemmli, U.K. (1970) Nature (London) 227, 248^254. G. (1989) Drug Metab. Rev. 20, 697^720. [30] Jez, J., Bennett, M., Schlegel, B., Lewis, M. and Penning, T. [2] Penning, T.M. (1993) Chem. Biol. Interact. 89, 1^34. (1997) Biochem. J. 326, 625^636. [3] Wang, X.P., Sheikh, S., Saigal, D., Robinson, L. and Weiner, H. [31] Hayes, J.D., Judah, D.J., Neal, G.E. and Nguyen, T. (1992) (1996) J. Biol. Chem. 271, 31172^31178. Biochem. J. 285, 173^180. [4] Leiber, C.S. (1994) J. Toxicol. Clin. Toxicol. 32, 631^681. [32] Hayes, J. et al. (1998) Chem.^Biol. Interact. 112, 51^67. [5] Maser, E. (1995) Biochem. Pharmacol. 49, 421^440. [33] O’Connor, T., Ireland, L., Harrison, D. and Hayes, J. (1999) [6] Li, R., Bianchet, M.A., Talalay, P. and Amzel, L.M. (1995) Proc. Biochem. J. 343, 487^504. Natl. Acad. Sci. USA 92, 8846^8850. [34] Hayes, J., Judah, D., McLellan, L., Kerr, L., Peacock, S. and [7] Hayes, J.D. and Pulford, D.J. (1995) Crit. Rev. Biochem. Mol. Neal, G. (1991) Biochem. J. 279, 385^398. Biol. 30, 445^600. [35] Grant, A., Sta¡as, L., Mancowiz, L., Kelly, V.P., Manson, [8] Jez, J. and Penning, T.M. (2001) Chem. Biol. Interact. 130^132, M.M., DePierre, J.W., Hayes, J.D. and Ellis, E.M. (2001) Bio- 499^525. chem. Pharmacol. 62, 1511^1519. [9] Jez, J., Flynn, T. and Penning, T. (1997) Biochem. Pharmacol. [36] Ja¡e, G., Murphy, J.E. and Robinson, O.P. (1976) Practitioner 54, 639^647. 216, 455^461. [10] Deyashiki, Y., Ohshima, K., Nakanishi, M., Sato, K., Matsuura, [37] Moss, J. and Bundgaard, H. (1992) Acta Pharm. Nord. 4, 301^ K. and Hara, A. (1995) J. Biol. Chem. 270, 10461^10467. 308.

FEBS 26272 5-7-02